计算机学报
計算機學報
계산궤학보
CHINESE JOURNAL OF COMPUTERS
2010年
4期
697-705
,共9页
网络安全%伪装攻击%入侵检测%shell命令%异常检测
網絡安全%偽裝攻擊%入侵檢測%shell命令%異常檢測
망락안전%위장공격%입침검측%shell명령%이상검측
network security%masquerade attack%intrusion detection%shell command%anomaly detection
伪装攻击是指非授权用户通过伪装成合法用户来获得访问关键数据或更高层访问权限的行为.近年来,伪装攻击检测在保障网络信息安全中发挥着越来越大的作用.文中提出一种新的用户伪装攻击检测方法.同现有的典型检测方法相比,该方法在训练阶段改进了对用户行为模式的表示方式,通过合理选择用户行为特征并基于阶梯式的序列模式支持度来建立合法用户的正常行为轮廓,提高了用户行为描述的准确性和对不同类型用户的适应性;在充分考虑shell命令审计数据时序特征的基础上,针对伪装攻击行为复杂多变的特点,提出基于多重行为模式并行挖掘和多门限联合判决的检测模型,并通过交叉验证和等量迭代逼近方法确定最佳门限参数,克服了单一序列模式检测模型在性能稳定性和容错能力方面的不足,在不明显增加计算成本的条件下大幅度提高了检测准确度.文中提出的方法已应用于实际检测系统,并表现出良好的检测性能.
偽裝攻擊是指非授權用戶通過偽裝成閤法用戶來穫得訪問關鍵數據或更高層訪問權限的行為.近年來,偽裝攻擊檢測在保障網絡信息安全中髮揮著越來越大的作用.文中提齣一種新的用戶偽裝攻擊檢測方法.同現有的典型檢測方法相比,該方法在訓練階段改進瞭對用戶行為模式的錶示方式,通過閤理選擇用戶行為特徵併基于階梯式的序列模式支持度來建立閤法用戶的正常行為輪廓,提高瞭用戶行為描述的準確性和對不同類型用戶的適應性;在充分攷慮shell命令審計數據時序特徵的基礎上,針對偽裝攻擊行為複雜多變的特點,提齣基于多重行為模式併行挖掘和多門限聯閤判決的檢測模型,併通過交扠驗證和等量迭代逼近方法確定最佳門限參數,剋服瞭單一序列模式檢測模型在性能穩定性和容錯能力方麵的不足,在不明顯增加計算成本的條件下大幅度提高瞭檢測準確度.文中提齣的方法已應用于實際檢測繫統,併錶現齣良好的檢測性能.
위장공격시지비수권용호통과위장성합법용호래획득방문관건수거혹경고층방문권한적행위.근년래,위장공격검측재보장망락신식안전중발휘착월래월대적작용.문중제출일충신적용호위장공격검측방법.동현유적전형검측방법상비,해방법재훈련계단개진료대용호행위모식적표시방식,통과합리선택용호행위특정병기우계제식적서렬모식지지도래건립합법용호적정상행위륜곽,제고료용호행위묘술적준학성화대불동류형용호적괄응성;재충분고필shell명령심계수거시서특정적기출상,침대위장공격행위복잡다변적특점,제출기우다중행위모식병행알굴화다문한연합판결적검측모형,병통과교차험증화등량질대핍근방법학정최가문한삼수,극복료단일서렬모식검측모형재성능은정성화용착능력방면적불족,재불명현증가계산성본적조건하대폭도제고료검측준학도.문중제출적방법이응용우실제검측계통,병표현출량호적검측성능.
Masquerade attacks are attempts by unauthorized users to gain access to confidential data or greater access privileges, while pretending to be legitimate users. Masquerade detection is now one of the major concerns of system security research. This paper proposes a novel method to distinguish legitimate users from masqueraders based on shell commands and multiple behavior pattern mining. In the method, behavioral patterns of legitimate users are characterized by shell command sequences of different lengths, and hierarchical sequence supports are employed to construct the normal behavior profiles of legitimate users, which improve the precision and adaptability of user profiling. In the detection stage, a model based on multiple sequence pattern parallel mining and multiple threshold joint decision is used to determine whether the monitored user's behavior is normal or anomalous. The model gives attention to both detection accuracy and computational efficiency, and is especially applicable for on-line detection. This study empirically demonstrated the promising performance of the method, and it has succeeded in getting application in practical host-based intrusion detection systems.